When conversion is disappointing, the default proposal is a redesign. It feels decisive and it is almost always the most expensive way to learn the least, because changing forty things at once tells you nothing about which of them mattered.
The alternative is unglamorous and considerably cheaper: measure where people actually leave, form one hypothesis about why, and change one thing.
Instrument before you touch anything
You cannot improve a number you are not recording. Before any change, the journey needs enough instrumentation that you can see each step and the drop between them.
- Define the conversion that matters — usually an enquiry or a booking, not a pageview.
- Record every step between arrival and that action, including form starts and abandons.
- Segment by device. Mobile and desktop frequently fail at different steps.
- Capture a baseline for at least one full business cycle before changing anything.
Find the leak, not the ugly bit
The step with the largest drop is where the money is, and it is often not where the design opinions are. A common finding is that the page everyone dislikes converts adequately and the form nobody looks at loses two thirds of the people who start it.
Watch the form specifically. Field count, error handling, and what happens on a failed submit account for a surprising share of lost enquiries, and none of them are visible in a screenshot.
One hypothesis at a time
A test needs a sentence: if we change X, then Y will improve, because Z. If you cannot write the “because”, you do not have a hypothesis, you have a preference.
- Change one variable per test where traffic allows it.
- Decide the success metric and the run length before you launch.
- Let it run to the end even when the early numbers look exciting.
- Record losing tests as carefully as winners — they are what stops the idea coming back next quarter.
Low traffic does not mean no testing
If volume is too low for a clean A/B split, sequential testing with a long enough window still beats guessing, and qualitative evidence carries more weight: session recordings, a single question on the thank-you page, or five minutes on the phone with someone who nearly did not enquire.
Keep the decision record
The most valuable artefact after a year of testing is not the conversion rate. It is the log of what was tried, what happened and what was decided, because it stops the organisation relitigating the same three ideas indefinitely.
Next step
AI A/B testing and conversion optimization follows exactly this order: instrument, find the leak, test one thing, write it down. If your traffic is healthy and your enquiries are not, tell us what should run better — we will start with the funnel data rather than a design opinion. You can also see how these systems fit together.